Setup Guide

Package Installation

1. Install a stable version of PyTorch (tested with version 1.13.1+cu116): https://pytorch.org/get-started/locally/
2. Install the latest version of MatPlotLib: https://pypi.org/project/matplotlib/
3. Install thop to track FLOPs and model parameter counts: https://pypi.org/project/thop/

Download and setup dataset

1. The HolStep dataset may be download here.
2. Create a new folder called data in the project root directory.
3. After download and extracing the HolStep dataset, place the holstep folder under the data folder.
4. you should have a folder called data with the following structure in the project root directory:

data
├── holstep
│   ├── train
│   │   ├── 01345
│   │   ├── .....
│   ├── test
│   │   ├── 01345
│   │   ├── .....

Train and evaluate the models

1. To train and evaluate the models, run the following command in the project root directory:

python epoch.py

2. We mainly train the SiameseCNNLSTM model and the SiameseTransformer models. To pick the specific model for faster completion, you can comment out the other models in the epoch.py file, under the if __name__ == '__main__': section.

3. You can choose to switch between cpu and gpu by changing the device variable in the top of the epoch.py file.

Refernces

Katz, Garrett. "TransformerHolstep.ipynb" Deep Automated Theorem Proving. CIS 700, Spring 2023.

Katz, Garrett. "TransformerMetamathV2.ipynb" Deep Automated Theorem Proving. CIS 700, Spring 2023.

Hunter, J. D. "Matplotlib: A 2D Graphics Environment." Computing in Science & Engineering, vol. 9, no. 3, 2007, pp. 90-95.

Harris, C.R., Millman, K.J., et al. Array programming with NumPy. Nature 585, 2020, pp. 357–362. https://doi.org/10.1038/s41586-020-2649-2

Paszke, A., Gross, S., et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library. Advances in Neural Information Processing Systems 32, Curran Associates, Inc., pp. 8024–8035. http://papers.neurips.cc/paper/9015-pytorch-an-imperative-style-high-performance-deep-learning-library.pdf

Zhu, Ligeng. "pytorch-OpCounter". GitHub, https://github.com/Lyken17/pytorch-OpCounter/